Update WRITEUP.md
Browse files- WRITEUP.md +1 -4
WRITEUP.md
CHANGED
|
@@ -6,7 +6,7 @@ code, compiled to real x86 machine instructions — executed by a computer whose
|
|
| 6 |
arithmetic and logic are **neural networks**, producing output **bit-identical**
|
| 7 |
to a normal CPU. Every pixel. Every byte of memory. Exact.
|
| 8 |
|
| 9 |
-
## The problem with neural networks (and why this should be
|
| 10 |
|
| 11 |
Neural networks are approximators. They're brilliant at "close enough" —
|
| 12 |
recognizing a cat, finishing your sentence. But a CPU lives or dies on
|
|
@@ -49,9 +49,6 @@ cpu_instrs: 11/11), and then a real cartridge game ran on it, with one full
|
|
| 49 |
frame replayed entirely through the neural units — bit-identical to a
|
| 50 |
conventional emulator, framebuffer and complete machine state.
|
| 51 |
|
| 52 |
-
****Then the obvious dare: a 1999 PC?** **
|
| 53 |
-
|
| 54 |
-
|
| 55 |
- **13 neural units** for the x86: 8-bit add/subtract slices carrying all five
|
| 56 |
x86 arithmetic flags, the logic ops, 1-bit shift slices, the multiplier
|
| 57 |
slice, and the instruction-decode tables. Each verified over its complete
|
|
|
|
| 6 |
arithmetic and logic are **neural networks**, producing output **bit-identical**
|
| 7 |
to a normal CPU. Every pixel. Every byte of memory. Exact.
|
| 8 |
|
| 9 |
+
## The problem with neural networks (and why this should be possible)
|
| 10 |
|
| 11 |
Neural networks are approximators. They're brilliant at "close enough" —
|
| 12 |
recognizing a cat, finishing your sentence. But a CPU lives or dies on
|
|
|
|
| 49 |
frame replayed entirely through the neural units — bit-identical to a
|
| 50 |
conventional emulator, framebuffer and complete machine state.
|
| 51 |
|
|
|
|
|
|
|
|
|
|
| 52 |
- **13 neural units** for the x86: 8-bit add/subtract slices carrying all five
|
| 53 |
x86 arithmetic flags, the logic ops, 1-bit shift slices, the multiplier
|
| 54 |
slice, and the instruction-decode tables. Each verified over its complete
|